What
sticks in my mind is the reduction of the number of surface temperature sites, with a claimed bias toward deleting locations
at higher altitudes and latitudes (i.e. those reporting
lower temperatures) yet leaving
historical reports from these locations in the average.